How a Roll of the Dice Can Improve Financial Planning

When you get to the point in your life when thinking about retirement becomes a real issue, the key question is: Will I run out of money before I die? The closer you get to that age, the more it becomes an obsession.

The typical planning approach is a linear one: You project a growth rate on your investments until the age you retire. Additional investments between now and then are calculated into the projection. Because you aren't sure what rates of growth you'll get, you run a couple of scenarios, say, 8% and 9% annual returns.

This approach has a number of inherent problems, which I'll get to in a minute. Fortunately, an alternative technique is gaining in popularity. It's called Monte Carlo Simulation (MCS), and it's is the methodology behind retirement calculation Web sites offered by Financial Engines (www.financialengines.com), T. Rowe Price (www.troweprice.com) and financialplanauditors.com (financialplanauditors.com).

In brief, instead of calculating two scenarios, MCS attempts to model thousands of random events and possibilities. The result is a much more realistic estimate of whether you'll be able to meet your retirement goals.

Let's look at how the linear approach works: Mr. Watkins is 50 years old and has $350,000. He plans to contribute $10,000 each year to his investment account for 14 years, at which time he'll retire. Then he plans to withdraw $90,000 each year. He is quite healthy. His contributions and withdrawals will be adjusted for inflation, which is assumed to be 3%.

Based on the commonly accepted way of calculating projections, let's assume growth rates of 8% and 9%. At 8%, Mr. Watkins would have about $1,270,000 at age 65, and at 9%, $1,429,000.

Mr. Watkins said he needed $90,000 a year. At 3% inflation each year, he would need about $136,000 a year at retirement to have the same buying power he has today. At the 8% growth rate, that means he would need to withdraw 10.7% a year of the $1,270,000. At 9%, it would be a 9.52% withdrawal per year from $1,429,000. Those are high rates of withdrawal and could result in his nest egg being depleted before he dies.

But it could be even worse. Let's say Mr. Watkins retired at the beginning of 1973, the start of a two-year bear market. The chart shows what would have happened to his nest egg. (The amounts shown assume he's withdrawing $136,000 a year and a 3% inflation rate.)

Year	Balance	Market Loss
1972	$1,429,000	--
1973	1,082,937	-14.7%
1974	658,125	-26.3%
1975	757,631	+37%

By 1975, the inflation-adjusted withdrawal from Mr. Watkins' investments amounts to almost 20% of the principle. If he lives longer than actuarial tables project, he'll most likely run out of money.

We may never again see a bear market like 1973-74. Then again, we might. But straight-line projections don't adequately account for the possibility.

Obviously, real life is not linear in the sense of market predictability and how long you'll live. The market goes up and down in varying degrees and will never do a straight 9%. Nor can you know for sure how long you'll live. Some people die years before the mortality tables say they should; some die years after.

"In order to plan for retirement you have to plan more realistically," says Rex Macey of Asset Allocation Simulation Software in Atlanta, maker of MCS software.

MCS takes into account time risk, investment risk and mortality risk. This technique of projection has been around for years. It was used mostly by statisticians, actuaries and engineers. In more recent years, a handful of financial planners have started using it for the most sophisticated and realistic retirement planning available.

How does it work? Ed McCarthy, a certified financial planner, offers this example in the November 1999 issue of the Journal of Financial Planning: "Assume that a stock's annual return could take only one of six distinct values each year: -20% , -5%, 0%, 5%, 20% and 30%. Each return is equally likely. In other words, each outcome has a 1/6 probability of occurring. If you wanted to simulate the stock's performance for a given year, you could take a die and assign a return to each side of the die. A roll of 1 is a -20% return, a roll of 2 represents a -5% return and so on. Each roll would simulate one annual return. If you rolled the die 10 times, you would have simulated one possible history of returns for that 10-year period. Because each roll produces a random result, your next 10 rolls should produce a different set of returns."

Sophisticated software can roll the die, so to speak, thousands of times to calculate the likelihood of thousands of possible investment returns, risk factors and life spans. It spits out an answer in percentage terms. In other words, the MCS suggests Mr. Watkins has a 54% probability of realizing his retirement goal.

That is not a high enough probability for success in retirement planning. Mr. Watkins will have to make some changes, such as retiring older, investing more, getting a higher return, reducing the income goal or a combination of these things. He should have more simulations run with these adjusted factors until he gets closer to a 90% probability of success.

Of course, if that doesn't work he could always go to the real Monte Carlo and throw the dice. But of course, who knows what he would be simulating with just one experience.

A question you might use to select a financial planner is: Do you use Monte Carlo Simulation for retirement planning? If not, you may want to use the "pass" line.